Late Holocene Climate and Glacier Fluctuations in the
Cambria Icefield area, British Columbia Coast Mountains
by Kate Johnson BSc, Lancaster University, 2008 A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of MASTER OF SCIENCE in the Department of Geography © Kate Johnson, 2010 University of Victoria All rights reserved. This thesis may not be reproduced in whole or in part, by photocopy or other means, without the permission of the author.Supervisory Committee
Late Holocene Climate and Glacier Fluctuations in the Cambria Icefield
area, British Columbia Coast Mountains
by Kate Johnson BSc, Lancaster University, 2008 Supervisory Committee Dr. Dan J. Smith, (Department of Geography) Supervisor Dr. Terri Lacourse (Department of Geography) Departmental MemberAbstract
Supervisory Committee
Supervisor Dr. Dan J. Smith, (Department of Geography) epartmental Member r. Terri Lacourse (Department of Geography) D D In the British Columbia Coast Mountains most dendroclimatological and dendroglaciological studies have focused on developing insights from tree‐ring sites located in the southern and central regions. By contrast relatively few studies have been conducted in the northwestern Coast Mountains, where exploratory studies reveal that significant climate‐radial growth relationships exist. The purpose of this study was to develop a proxy record of climate change from tree rings and to reconstruct the late Holocene glacial history of two outlet glaciers spilling eastward from the Cambria Icefield. Dendroclimate investigations were conducted using mountain hemlock (Tsuga mertensiana) trees growing on three high‐elevation montane slopes. The three stands located along a 35 km transect cross date to form a master chronology for the region spanning 409 years (1596 to 2007 A.D.). Correlation analyses show that the radial growth of the regional tree‐ring chronology corresponds to variations in the mean June‐July‐August (JJA) air temperature. The relationship between the two variables was used to reconstruct mean JJA air temperature from 1680 to 2007 A.D.). The reconstruction illustrates warm and cool intervals that are synchronousto those derived from other paleoenvironmental research in this region. The proxy record also highlights annual to inter‐decadal climate variability likely resulting from atmospheric‐ocean circulation patterns described by the El Niño‐Southern Oscillation and the Pacific Decadal Oscillation. The late Holocene behaviour of White and South Flat glaciers was investigated using radiocarbon dating techniques, dendrochronological cross‐dating techniques and geomorphological analysis of sedimentary units within the White and South Flat glacier forefields. Evidence for a First Millennial Advance (FMA) cumulating around 650 A.D. and early Little Ice Age (LIA) advances at 1200 and 1400 A.D. were documented. These advances are contemporaneous with the late Holocene activity of glaciers throughout the region, suggesting coherent broad‐scale climate forcing mechanisms have influence glacial mass balance regimes over at least the last two millennia. The dendroclimatological and dendroglaciological findings of the study provide the first annually‐resolved climate record for the region and help to enhance our understanding of late‐Holocene glacier behaviour in the Cambria Icefield Area. The thesis documents the complex interactions between climate and the radial growth of mountain hemlock trees in the Pacific Northwest, and describes the role that long‐term climate variability played in glacier dynamics during the FMA and LIA.
Table of Contents
Supervisory Committee ... ii Table of Contents ... v Abstract ... iii List of Tables ... vii List of Figures ... viii Acknowledgments ... xii Chapter 1: Introduction ... 1 1.1 Introduction ... 1 1.2 Research Objectives ... 3 1.3 Thesis Format ... 3 Chapter 2: A 323year dendroclimatic reconstruction of summer mean air temperature in the northern British Columbia Coast Mountains, Canada ... 5 2.2 2.1 Introduction ... 5 Research Background... 6 2.32.2.2 Dendroclimatology of mountain hemlock ... 9 Location ... 9 2.42.3.1 Study Sites ... 10 Methodology ... 12 2.4.1 Dendrochronology ... 12 2.52.4.2 Climate Data ... 15 Results ... 17 2.5.1 Tree‐ring chronologies ... 17 2.62.5.2 Climate station analysis ... 20 Analysis ... 23 2.6.1 Climate‐Growth Responses ... 23 2.6.2 Dendroclimate Reconstruction ... 26 2.6.3 PDO and ENSO Circulation Patterns ... 28 2.7 Discussion ... 31 2.8 Summary ... 34 Chapter 3: Dendroglaciological reconstruction of late Holocene glacier activity at White and South Flat Glaciers, Boundary Range, northern British Columbia Coast Mountains ... 35 3.2 3.1 Introduction ... 35 Research Background... 36 3.2.1 Tree‐rings and dendroglaciology ... 36 3.33.2.2 Holocene Glacial Activity in the British Columbia Coast Mountains ... 38 Research Area ... 40 3.43.3.1 Study Sites ... 43 Research Methods ... 49 3.4.1 Sample Preparation and Measurement ... 50 3.5 Observations ... 513.5.1 Dendrochronology ... 51 3.5.2 Dendroglaciology ... 53 3.5.2.1 Site 1 ... 53 3.5.2.2 Site 2 ... 55 3.5.2.3 Site 3 ... 58 3.5.2.4 Site 4 ... 60 3.5.2.5 Site 5 ... 60 3.5.2.6 Site 6 ... 62 3.5.2.7 Site 7 ... 63 3.5.2.8 Site 8 ... 64 3.63.5.2.9 Site 9 ... 66 Results ... 67 3.6.1 Interpretation ... 69 3.7 Discussion and Synthesis ... 75 3.8 Summary ... 79 hapter 4: Conclusions ... 81 eferences ... 84 C R
List of Tables
Table 2.1: Tree‐ring chronology statistics... 17 able 2.2: Correlation matrix showing interseries correlations between site hronologi 20 T c es and master chronology. ... able 2.3: Correlations between instrumental climate variables and the residual egional ch 4 T r ronology. Significance level = 0.01. ... 2 able 3.1: Estimated historical terminus retreat rates at the White Glacier and the outh Flat T S Glacier. ... 43 able 3.2: ... 52 T Living tree‐ring chronology statistics. ... able 3.3: Subfossil study sites at White Glacier and their associated UTM oordinate 53 T c s and elevation values. ... able 3.4: Summary of conventional age radiocarbon dated samples from White nd South ... 55 T a Flat Glaciers. ... Table 3.5: The floating tree‐ring chronologies created for this thesis and their conventional age dates assigned after unsuccessful cross‐dating into the living tree‐ ing chronology. If the sample was used in the FMA Chronology (Figure 3.18), then he perimeter age date is recorded. ... 56 r tList of Figures
Figure 2.1: Location of the sample sites and climate stations used in this thesis. ... 7 igure 2.2: The Windy Pass Site at 1080 m asl, with a Remote Avalanche Weather tations (RA .... 11 F S WS) used for avalanche prediction in the region. Photo: Dan Smith. Figure 2.3: The South Flat Site located west of South Flat Glacier at 1035 m asl. rees were growing on a south‐east facing slope dominated by mountain hemlock rees and sc T t attered with subalpine fir. Photo: Dan Smith... 12 igure 2.4: The residual tree‐ring chronologies for (a) Surprise, (b) Windy Pass, and c) South Fl F ( at sites with a 10 year running mean. ... 18 Figure 2.5: Living mountain hemlock chronology displayed with a five‐year running mean. The number of series included in the chronology (sample depth) is shown at he bottom of the graph. The master chronology contains 27 series; the series ntercorrela t i tion value is 0.544 and the average mean sensitivity is 0.244. ... 21 igure 2.6: A comparison of annual mean air temperature between Stewart Airport nd Dease L F a ake. ... 22 igure 2.7: A comparison of annual mean air temperature between Stewart Airport nd Dease L F a ake. ... 23 Figure 2.8: Coefficients calculated between the mountain hemlock chronology and the mean temperature time series from Dease Lake using DENDROCLIM2002. The lack bars represent correlation coefficients; the gray bars represent response unction co ... 24 b f efficients. Months in block letters are from the previous year. ... Figure 2.9: Scatterplot comparing the radial growth of mountain hemlock and verage June‐July‐August temperature at Dease Lake over the instrumental record 1943 – 200 a ( 3). ... 25 Figure 2.10: This figure illustrates bootstrapped moving interval response function analyses calculated using the program DENDROCLIM2002. Forty‐eight year intervals were employed to test the strength of relationships between tree ring indices and the Dease Lake monthly June to August air temperature values over time. The panels depict the relationships between the mountain hemlock chronology and mean temperature values. Statistically significant positive elationships are highlighted in shades of red and negative relationships in shades f blue. Months of the previous year are identified with capital letters. ... 26 r oigure 2.11: June‐July‐August temperature model based on a linear relationship ith the rad 7 F w ial growth of mountain hemlock. ... 2 igure 2.12: Reconstructed precipitation anomaly record for the Cambria Icefield. ... ... 28 F . ... Figure 2.13: Tree ring index – June‐July‐August sea surface temperature correlation. The tree rings have a positive relationship (r = 0.2 to 0.3) to sea surface temperatures across the equator, along the South American coast to 30˚ south, and ff the coast of northern BC and Alaska. There is a negative correlation (r = ‐0.2 to ‐ .4) with sea o 0 surface temperature in the central north Pacific. ... 29 Figure 2.14: (a) Summer Temperature Anomalies. (b) The wavelet power spectrum. The contour levels are chosen so that 75%, 50%, 25%, and 5% of the wavelet power is above each level, respectively. Black contour is the 5% significance level, using a white‐noise background spectrum. (c) The global wavelet power spectrum (black ine). The dashed line is the significance for the global wavelet spectrum, assuming he same si l t gnificance level and background spectrum as in (b). ... 30 igure 3.1: Location of the Cambria Icefield and key glaciers within the vicinity and eferred to F r in the text. ... 37 igure 3.2: Site map of the White Glacier and South Flat valley, showing the location f the glacie .. 42 F o rs and study sites. ... igure 3.3: Aerial photograph from 1991. The white line demarcates White and outh Flat g 4 F S laciers’ maximum upvalley extent during the LIA (see Figure 3.21). ... 4 Figure 3.4: The White Glacier forefield, looking from White Glacier towards White ake. The White Glacier forefield is heavily colonised by sika alder (Alnus sinuata). ... .... 45 L . ... Figure 3.5: The view from White Glacier in late‐July 2007. In the foreground is showing the location of a recently drained glacially damned lake. This may orrespond to the ice contact lake shown in Figure 3.6 (site 3), noted by Groves 1946). ... ... 46 c ( ... igure 3.6: Geological map from Groves (1973). 1. White Glacier; 2. South Flat lacier; 3. A ... 48 F G small lake, ponded by White Glacier, and; 4. White Lake. ... igure 3.7: The view of South Flat valley and South Flat river. The white box ighlights h raded delta sedim .. 49 F h ighlights the location of late‐LIA prog ents. ... Figure 3.8: Living tree‐ring mountain hemlock (Tsuga mertensiana) chronology from the South Flat Glacier. ... 52
Figure 3.9: Looking north towards site 1. Three units are highlighted: a 10 m thick matrix‐supported till (1A), wood samples taken from the mat (Sample 01_01) date to 605 A.D. This unit is overlain by a 7 m thick till unit (1B) capped by an orange‐ tained sedimentary horizon. Resting on the surface of this horizon is a 7 m thick till nit (1C) wh 4 s u ose upper surface features small north‐west trending glacial flutes. .... 5 igure 3.10: Looking south towards site 2. Two white boxes highlight the location f the two po ... 58 F o ckets of sampled wood with sample 02_03 dated to 510 A.D. ... Figure 3.11: East towards the exposed wood at site 3 against bedrock. Samples were located at the base of a 10m high bedrock outcrop in a diamicton intermixed with partially buried and buried bole segments up to 6m in length. Two radiocarbon ges were assigned to this site: 640 A.D. (Sample 03_10; within white box) and 515 .D. (Sample a A 03_14; not shown on photograph). ... 59 Figure 3.12: Looking south across the exposed debris at site 5. The gully ran for 20 north to south and was 6 m deep. Sample (05_01) indicates these trees were illed ca. 625 m k A.D. ... 61 Figure 3.13: Looking north from White Glacier to site 6. Wood was sampled from etrital wood in an alder gully. The white box shows the location of the 5 samples ogs killed ca ... 62 d l . 640 A.D. ... Figure 3.14: Looking south towards White Glacier at Site 7. The white box highlights the in situ Sample 07_01 (765 A.D.). Disks were taken from samples ithin the river channel and in an adjacent deposit to the west of the river channel o form Chro . 63 w t nologies 7‐1 and 7‐2... Figure 3.15: The steeply‐dipping laminated sands and gravels capped by a till at ite 8. Sample 08_11 indicates than the trees in this deposit were killed ca. 1080 A.D. ... . 65 s . ... igure 3.16: Bedded fine‐grained rythmites at site 8 located beside a prominent edrock gorg F b e 900 m from the 2007 terminus of South Flat Glacier. ... 65 Figure 3.17: Sample 9_16 at site 9 (lower), radiocarbon dated to ca. 1220 A.D. Site 9 consisted of three woody horizons, at 818 m asl (lower), 851 m asl (middle; dated to 1355 A.D. and a third horizon (upper) that was inaccessible at around 900 m asl. hese three woody horizons appear to be matched with equivalent units visible on he west sid ... 66 T t e of the valley, highlighted with three white lines. ... Figure 3.18: First Millennial Advance Chronology anchored by conventional radiocarbon date from sample 03_10 (Table 3.5). The chronology consists of samples from the White Glacier forefield, South Flat Valley and the Todd Icefields (from Jackson et al. [2008]). The chronology extends from 251 – 664 A.D. (393 years) and consists of 38 tree‐ring records. ... 68
Figure 3.19: The advance of White Glacier killed this tree (within the white box) rior to 486 A.D. As White Glacier advanced into and overwhelmed a standing orest, it push .. 70 p f ed the tree against this bedrock. The tree was exposed after 2001. .. Figure 3.20: A schematic diagram of the ice surface profile of White Glacier: (A) uring the FMA advance at 515 A.D., based on the cross‐dating of detrital wood from ite 4; and (B es. 72 d s ) during the late LIA, based on the location of the terminal morain Figure 3.21: Schematic of the expansion of White Glacier and the subsequent interaction of White Glacier with South Flat Glacier between ca. 600 A.D. to the early Little Ice Age. In panel 1, the glacier is advancing to site 3 with South Flat river flowing beneath the glacier. In panel 2, the glacier has thickened sufficiently to force the ponding of a lake, killing sample 07_01 at site 7. South flat river subsequently flows around White Glacier. Panel 3 depicts the advance of South Flat Glacier, causing the formation of a prograding delta, deposition of sediments and death of trees at site 9. White and South Flat glacier are not confluent until the late Little Ice ge (panel 4), where all of the South Flat valley and White Glacier forfield is covered ith glacial i 3 A w ce. ... 7 Figure 3.22: Looking up the South Flat valley. Sample 07_01 (765 A.D.) from site 7 is highlighted with a white box. Site 9 encompasses three woody horizons: The lower horizon (1220 A.D.; sample 09_16), middle horizon (1355 A.D.; 09_18), and an ndated upper horizon. These three woody horizons appear to be matched with quivalent units vis u e ible on the west side of the valley (see Figure 3.17). ... 74 Figure 3.23: The wavelet power spectrum for the (a) Live Chronology, and; (b) the Stewart Chronology. The contour levels are chosen so that 75%, 50%, 25%, and 5% of the wavelet power is above each level, respectively. The cross‐hatched region is the cone of influence, where zero padding has reduced the variance. Black contour is the 15% significance level, using a white‐noise background spectrum. (c) The global wavelet power spectrum (black line). The dashed line is the significance for the global wavelet spectrum, assuming the same significance level and background spectrum as in (b). ... 77
Acknowledgments
First and foremost, to Dan Smith: your patience with my “early” rising (at 9am) in the field, guidance down Jambeau Glacier’s crevasses, enthusiasm for grizzly bears in the Coast Mountains, and love for earthquakes in the 1700s has been second to none. Those sleepless nights wondering if I had a granola bar in my bag couldn’t have been completed without you! Thank you also to Terri Lacourse, whose speedy edits have improved this hesis in ways I couldn’t have achieved, even after the 100th attempt. t The field work couldn’t have been completed without the Midnight Dream and Knight Rider gangs. Thanks for all the hazy memories involving Hyderizing, karaoke, KAJOs and pie (man, I love me some pie). And thank you to all the lovely adies I met back at the “VicTree” lab. l To my friends, whether in the UK or Canada, I appreciate your support throughout this adventure, and to my family, for your understanding on why I was ,000 miles away and still complaining about the Northampton Town score. 5 Last but by no means least, I thank kypp; I’ll be indebted to your patience and attempt at understanding my frustrations over cross dating and statistics. For this, I probably owe you a few dinners at the Ramore. And also for those immortal words: “wait, so, Dendrochro‐ whatsit is a real word?”Chapter 1
Introduction
1.1 Introduction
The global climate is undergoing substantive changes that are resulting in fundamental shifts in climate patterns which could significantly impact our society and ecosystems (Mann et al. 1999). To fully understand the nature and character of these changes, it is essential that we develop an understanding of past climate conditions and place these historical changes into a longer temporal context (Hansen and Lebedeff 1987). Due to limitations in the length and distribution of instrumental climatic records, high‐resolution multi‐proxy methods have proven critical in providing the insights needed to interpret current climate trends and to establish long‐term ranges of variability (Black et al. 2009). Prior and ongoing research suggests that tree‐rings provide the kind of annually resolved insights that are necessary for extending climatic records back several millennia (D’Arrigo et al. 2005). The science of dendrochronology (or tree‐ring studies) has focused on the annual growth rings of temperate zone trees to establish the date and chronological order of past events (Fritts 1976). Tree‐ring chronologies can be dendroclimatically interpreted to reconstruct a variety of environmental variables when correlated with instrumental data or records (Cook and Kairiukstis 1990). While long‐lived tree species provide annually‐resolved chronologies that can extend back as far as 3,000 years, subfossil wood samples have been employed toextend tree‐ring and proxy climate records back through the Holocene (Grudd et al. 2002). Dendroclimatology is the science that uses tree‐rings to study present climate and reconstruct past climate (Luckman 2007). This methodology allows dendrochronologists to place existing short instrumental climate records in a longer temporal context, and allows for comparison of the spatial patterns of climate change. In Pacific North America Holocene regional‐scale climate changes are known to directly impact upon the mass balance state of glaciers resulting in sustained periods of expansion and retreat (Reyes et al. 2006). Dendroglaciology is a branch of dendrochronology that uses tree‐rings to study and date the response of glaciers to changing climates (Luckman 1986). It allows for the development of long‐term tree‐ring chronologies for dating events over large areas, and hence for examination of the synchronicity of glacier activity at different time scales (Luckman and Villalba 2001). Applied dendroglaciological research has successfully established a link between glacier mass balance and tree‐ ring‐width variability (Smith and Lewis 2007a). In the Coast Mountains of British Columbia (B.C.), prior dendroclimatological and dendroglaciological studies have primarily focused on developing insights at tree‐ring sites located in the southern and central regions (e.g., Larocque and Smith 2003; Allen and Smith 2007). By contrast relatively few tree‐ring studies have been conducted in the north Coast Mountains, where exploratory studies by Penrose (2007) and Jackson et al. (2008) revealed that significant climate‐radial growth relationships do exist. The purpose of this study
was to reconstruct the environmental history of the Cambria Icefield area. Specific attention was given to developing a proxy record of climate change and to reconstructing the late Holocene glacial history of two outlet glaciers spilling eastward from the icefield. The overarching intent was to determine whether the late Holocene climate and glacier fluctuations in this regional are similar to those recorded in the central and southern Coast Mountains.
1.2 Research Objectives
The research had three specific objectives: 1. to develop a proxy record of climate change from tree ring‐width chronologies collected in the vicinity of the Cambria Icefield. 2. to describe the late Holocene glacial history of White and South Flat glaciers in the Cambria Icefield. 3. to investigate the long‐term glaciological response of White and South Flat glaciers to changing climates .1.3 Thesis Format
Chapter 2 presents the findings of a dendroclimatological reconstruction from tree‐ring records collected at sites in the vicinity of the Cambria Icefield. Chapter 3 describes the dendroglaciological findings from field investigations completed at White and South Flat glaciers. Chapters 2 and 3 are formatted as manuscripts prepared for submission to refereed journals. Chapter 4 provides an overview of the findings of the research, and considers the relationship between long‐term climate changes and glaciological behaviour of the White and South Flatglaciers. The chapter concludes with a presentation of the limitations study and provides recommendations for future research.
Chapter 2
A 323year dendroclimatic reconstruction of summer mean air
temperature in the northern British Columbia Coast Mountains,
Canada
2.1 Introduction
Climate plays an important role in limiting the annual radial growth of montane trees in the British Columbia (B.C.) Coast Mountains (e.g., Gedalof and Smith 2001a; Larocque and Smith 2004; Parish and Antos 2006). Previous studies in the region highlight a correlation between apical growth and environmental factors such as summer growing season temperature, previous growth year temperature, and spring snowpack depth (Larocque and Smith 2004). The year‐to‐ year correlation of tree rings to these seasonal climate variables indicates these trees have considerable potential for constructing dendroclimatic proxy records (Luckman 2007). In Pacific North America (PNA) mountain hemlock (Tsuga mertensiana [Bong.] Carr.) trees are widely dispersed, covering a latitudinal range extending from southern Alaska to northern California (Krajina 1969; Means 1990). In B.C. the mountain hemlock zone occupies elevations between 900 to 1800 m asl in the south (lower on windward slopes, higher on leeward slopes), and 400 to 1000 m asl in the north (Pojar et al. 1987). Dendroclimatological analyses show that mountain hemlock trees are sensitive recorders of annual changes in high elevation climates (Laroque and Smith 2001). Researchers have shown that theirradial growth is conditioned by a limited number of climate stressors depending upon their location (Graumlich and Brubaker 1986; Gedalof and Smith 2001b). These climate‐radial growth relationships have been used to construct proxy models of past climates using both single chronologies and regional networks of chronologies (i.e., Wiles et al. 1998; Gedalof and Smith 2001b). The goal of the research presented in this chapter was to construct an annually‐resolved proxy record of climate change in the northwestern Coast Mountains. The intent was to quantify the climate‐radial growth associations displayed by living mountain hemlock stands found growing in close proximity to the Cambria and Todd icefields. Historic climate station data was compared to the total ring‐width growth trends over the period of record and used to construct a proxy record of climate extending back over the extent of the tree‐ring record. The proxy reconstruction was compared to several climate indices to reveal which forcing mechanisms may have interacted to define the regional climate over the last 300 years.
2.2 Research Background
2.2.1 Climate forcing in the Pacific Northwest The Coast Mountains flank the western coastline of B.C. (Figure 2.1). This high mountain landscape extends for ca. 1600 kms from the northwestern corner of the province southward to close to the international border with Washington State (Means 1990). The climate of the region is moderated by proximity to the Pacific Ocean (Mantua et al. 1997), with inter‐ annual to inter‐decadal variabilityFigure 2.1: Location of the sample sites and climate stations used in this thesis.
largely resulting from atmospheric‐ocean circulation patterns described by the El Niño South Oscillation (ENSO) and the Pacific Decadal Oscillation (PDO; Bonsal et al. 2001). ENSO describes the influence of both ocean and atmospheric circulation patterns in PNA (Rasmussan and Wallace 1983). It is widely accepted as representative of the climate forcing parameters that lead to hemispheric‐scale inter‐annual climate variations (Mantua and Hare 2002). Periodic shifts in the strength of ENSO result in weather and climate anomalies every 7 to 10 years (Rasmussan and Wallace 1983). In the Coast Mountains, ENSO events are generally correlated to anomalously warmer and drier winter months (Bonsal et al. 2001). The PDO is a long‐term El Niño‐like ocean and atmospheric circulation pattern that occurs in the Northern Pacific, with a positive (or warm) phase or a negative (or cool) phase (Manuta and Hare 2001). It is known to change phases with an abrupt step‐like shift in mean winter sea level pressure (Trenberth 1990; Mantua and Hare 2002). Climate anomalies associated with positive (or warm) phases coincide with anomalously warmer and wetter periods in PNA (Mantua and Hare 2002). Three PDO shifts in the past century resulted in “cool” phases from 1890 to 1924, and again from 1947 to 1976; and, “warm” phases from 1925 to 1946 and from 1977 to the mid‐1990s (Mantua et al. 1997). Tourre et al. (2001) indicate that the PDO has an interdecadal mode of between 12 to 25 years and a decadal mode of between 9 to 12 years. Gedalof and Smith (2001b) employed dendrochronological methods to construct a proxy record of spring PDO conditions
over the last 400 years. Their reconstruction illustrates the long‐term average duration between PDO shifts is 23 years. 2.2.2 Dendroclimatology of mountain hemlock Previous research established that mountain hemlock trees are a useful species for dendroclimatic studies due to their spatial range and sensitivity to climate (Gedalof 2002; Penrose 2007). Generally, mountain hemlocks are found at locations characterized by cool, wet climates and deep winter snowpacks (Peterson and Peterson 2001). Annual radial growth increment of mountain hemlock has been correlated with mean summer temperature (Gedalof and Smith 2001a), winter precipitation described by spring snowpack depth (Graumlich and Brubaker 1986; Gedalof and Smith 2001a; Peterson and Peterson 2001) and April precipitation totals (Penrose 2007).
2.3 Location
Sampling was undertaken at sites located in close proximity to the Todd and Cambria icefields (Figure 2.1). This glaciated high mountain region is characterized by topographic relief exceeding 900 m (Figure 2.1). The local tree line is found at ca. 1500 m asl, significantly above the present‐day terminus of glaciers in the region at ca. 1200 m asl (Laxton 2005; Jackson et al. 2008). Montane forests in this region of the Coast Mountains consist of single‐ species stands of mountain hemlock or co‐dominant stands of mountain hemlock and subalpine fir (Abies lasiocarpa [Hook.] Nutt.). Mountain hemlock foreststypically have a dense shrub growth under the tree canopy, with thick carpet of moss covering the forest floor (Klinka et al. 1991). Adjusted and Homogenized Canadian Climate Data (AHCCD) data indicate the montane forests in the vicinity of Stewart are characterized by warm, wet conditions with mean annual temperatures of 5.6°C, and annual total precipitation of 1458 mm(Environment Canada 2009). The eastern slopes of the Coast Mountains in the vicinity of the Cambria and Todd icefields experience drier and cooler conditions with mean annual temperatures of ‐0.9°C, and annual total precipitation of 292 mm (Environment Canada 2009). 2.3.1 Study Sites Samples were collected from three mountain hemlock stands found along a 35 km north to south transect (Figure 2.1). The sampling sites are located close to the easternmost extent of mountain hemlocks in the region (Pojar et al. 1987; Klinka et al. 2001). The northernmost site is located in the headwaters of Surprise Creek at 910 m asl (Lat 56° 12’ N, Long 129° 36’ W; Figure 2.1). Positioned on a moderately sloped forested ridge line between the recently deglaciated forefields of Surprise Glacier (unofficial name) to the south and an unnamed glacier to the north, the mature mountain hemlock trees (<6 m tall) found at the site were intermixed with mature subalpine firs whose maximum age exceeded 355 years (Jackson et al. 2008).
The Windy Pass site (Lat 56° 06’ N, Long 129° 30’ W; Figure 2.1) is located 8.6 km east of Bear Pass above the Stewart‐Cassier Highway (Highway 37) close to the upper altitudinal extent of standing trees. Located on a southerly‐facing minor bench at 1090 m asl, 11 km southeast of the Surprise site, the forest at Windy Pass site is dominated by short stature (<4 m) mature mountain hemlock trees (Figure 2.2). Figure 2.2: The Windy Pass Site at 1080 m asl, with a Remote Avalanche Weather Stations (RAWS) used for avalanche prediction in the region. Photo: Dan Smith. The South Flat site (Lat 55° 49’ N, Long 129° 29’; 1035 m asl; Figure 2.1) is located on a steep southeast‐facing forested site at 1035 m asl in close proximity to White Glacier (unofficial name), approximately 33 km south of the Windy Pass site.
The site is dominated by a mature even‐aged mountain hemlock forest (>6 m eight) and contains scattered subalpine fir (Figure 2.3). h Figure 2.3: The South Flat Site located west of South Flat Glacier at 1035 m asl. Trees were growing on a south‐east facing slope dominated by mountain hemlock trees and scattered with subalpine fir. Photo: Dan Smith.
2.4 Methodology
2.4.1 Dendrochronology Tree ring samples were collected from mature trees with no obvious signs of crown damage or butt rot. Samples were collected at breast height with 18” increment borers using standard dendrochronological techniques (Stokes and Smiley 1964). Two increment cores ≥90° apart were extracted from each tree topith and stored in plastic straws for transport to the University of Victoria Tree Ring Laboratory. The cores were allowed to air‐dry, glued into slotted mounting boards, and sanded with progressively finer sandpaper to a 600‐grit polish (Stokes and Smiley 1968). Ring widths were measured to the nearest 0.01 mm along a single radius using WinDendro software and a high resolution flatbed scanner (Guay et al. 1992). For series with exceptionally narrow annual rings, the ring widths were measured with MeasureJ2X software (VoorTech Consulting 2009) to the nearest 0.01 mm using a Velmex‐stage equipped with a microscope, CCD video display and Metronics QC‐10V digital readout. Each ring‐width series was first visually cross dated by comparing narrow marker rings with CDendro software (Cybis Elektronic & Data AB). Following this the International Tree Ring Database (ITRDB) software program COFECHA was used to verify the cross dating using block correlations between each tree series and the master chronology (Holmes et al. 1986). COFECHA correlations were calculated using a 50‐year segment length lagged successively by 25 years at a one‐ tailed 99% confidence level (Grissino‐Meyer 2001). Ring width series with anomalous growth patterns not significantly correlated to the group were removed from the data set. A regional master chronology was created by combining and cross dating the three individual chronologies. Any ring width series that did not significantly correlate to the master chronology was removed from the data set to allow for the strongest possible series intercorrelation.
All of the ring width series within the master chronology were transformed into stationary, dimensionless indices to remove growth trends related to tree age and stand dynamics (Cook 1985). The indices were established using double detrending options in the IITRDB software program ARSTAN (Cook and Holmes 1986) to retain any low‐frequency variability within the resulting chronology (Cook et al. 1990). Initially a best‐fit negative exponential curve or linear trend line was fitted to each individual series. Each observed ring width in the series was then divided by this value. Following this, the individual cores were detrended a second time by fitting a cubic smoothing spline with a 50% frequency cutoff of 95 years (Cook and Peters 1981). To create a single dimensionless value for each growth year, each ring‐width‐index series was prewhitened using autoregressive and moving average models to remove any autocorrelation effects (Cook 1985; Biondi and Swetnam 1987). A robust mean was then used to combine the etren d ded series from each site into residual master site chronology (Fritts 1976). To ensure robust signal strength through time, each site chronology and the master chronology was terminated at an established “expressed population signal” (EPS) value (Wigley et al. 1984). EPS values were calculated for 25‐year moving periods to quantify signal strength through time; the cut‐off year was defined as the central year within the last 25‐year segment where the EPS value was greater than the 0.85 cut‐off value (Wilson and Luckman 2002). Bootstrapped correlation and response function analyses were undertaken to evaluate the climate‐radial growth response of the chronologies using the software program DENDROCLIM2002 (Biondi and Waikul 2004). Climate variables
were compared to the residual chronologies using bootstrap correlations. The stability of the climate influence on radial growth was analysed by calculating block correlations of 30 years duration with an overlap of 15 years. After determining which climate variable was significantly correlated to the master chronology, a linear regression model was used to represent the proxy relationship between ring width and climate using the most recent 50% of the data (LeBlanc and Terrell 2001; Hughes 2002). A split period verification analysis and a simple Pearson's correlation were used to verify the linear regression model’s ability to represent the remaining 50% of historic data (Fritts 1976). Following this procedure, the model was invoked over the duration of the residual master chronology to illustrate climate variability through time. 2.4.2 Climate Data Mean, maximum, and minimum monthly temperature and precipitation data were obtained from the AHCCD website (http://www.cccma.ec.gc.ca/hccd/). Missing values were replaced with long‐term monthly averages. Only two climate stations in the region have lengthy historical records. The closest to the sampling sites is located at the Stewart Airport (Climate Station 1067742; Lat 55° 56′ N, Long 129° 59′ W; Figure 2.1). Located on the windward side of the Cambria Icefield at 7 m asl, the station experiences a coastal maritime climate and has been operational since 1911. The annual air temperature averages 5.6°C, with the coldest month (January) averaging –3.7°C and the warmest (July)
15.1°C (Environment Canada 2009). The total annual precipitation at Stewart averages 1843 mm, with more than 30% of this falling as snow. The nearest climate station located on the leeward side of the Boundary Range is found at Dease Lake, approximately 260 km northeast of the Surprise site (Climate station 1182285; Lat 58° 43’ N; Long 130° 02’ W; Figure 2.1). The station is located at 807 m asl and has been operational since 1944. The annual air temperature averages ‐0.9°C, with the coldest month averaging ‐14.2°C and the warmest 12.2°C (Environment Canada 2009). Historical sea surface temperature (SST) records provide insights into the role that climate forcing events associated with the PDO and ENSO have on radial growth (D'Arrigo et al. 1999). The Hadley Centre’s HADISST’s global gridded 1° Latitude x 1° Longitude mean monthly SST from 1870 to present were used as a surrogate for PDO and ENSO phases. This data was correlated with the master chronology using the interactive KNMI climate explorer website (http://climexp.knmi.nl; van Oldenborgh and Burgers 2005). To understand the temporal rhythm of these climate forcing events on the radial growth of mountain hemlock trees, an interactive web‐based wavelet analysis1 was employed. In this instance, a Gaussian 2 function was used as the base function with a 5% white noise reduction. 1http://paos.colorado.edu/research/wavelets; (Torrence and Compo 1998)
2.5 Results
2.5.1 Treering chronologies Twenty‐six ring width series from 20 trees were included in the chronology from the Surprise site. The final chronology spans 409 years (1596 to 2007), with an EPS cut off at 1680. The ring width series has a mean inter‐series correlation of 0.535 and an average mean sensitivity of 0.211 (Table 2.1). Table 2.1: Tree‐ring chronology statistics. Thirty‐six ring width series from 23 individual trees were included in the final analysis of increment cores collected at the Windy Pass site. The site chronology spans 391 years (1613 to 2004), with an EPS cut off at 1730 due to a limited sample depth beyond this. The ring width series have a mean inter‐series correlation of 0.595 and a mean sensitivity of 0.271 (Table 2.1).Site Latitude Long tudei Elevation (m asl) Series mean first‐order autocorrelationa
Series mean sensitivityb Total series length (years) Surprise 56° 12’ 129° 36’ 910 0.535 0.211 408 Windy Pass 56 ° 06’ 129° 30’ 1090 0.595 0.271 391 South Flat 55°49’ 129° 29’ 10 5 3 0.548 0.231 373 Master ‐ ‐ ‐ 0.544 0.244 411 a a measure of the strength of the signal (typically the climate signal) common to all sampled trees at the site (Grissino‐Mayer 2001) b a measure of the relative change in ring‐width from one year to the next in a given series (Grissino‐Mayer 2001)
Twenty‐seven ring width series from 17 individual trees were included in the final analysis of samples from the South Flat site, the site chronology spans 373 years (1635 to 2007), with an EPS cut off at 1700. The ring width series have a mean inter‐series correlation of 0.548 and a mean sensitivity of 0.231 (Table 2.1). The site chronologies strongly correlate to one another (Figure 2.4). While he radial growth trends between the South Flat and Windy Pass sites are strongly t F a igure 2.4: The residual tree‐ring chronologies for (a) Surprise, (b) Windy Pass, nd (c) South Flat sites with a 10 year running mean. 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1600 1650 1700 1750 1800 1850 1900 1950 2000 Year R esid u al R in g -wid th Surprise
10 per. Mov. Avg. (Surprise)
(a)
0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 (b) 2 1600 1650 1700 1750 1800 1850 1900 1950 2000 Year R esidu al Rin g -wid th Windy Pass
10 per. Mov. Avg. (Windy Pass)
0.4 0.6 0.8 1 1.2 1.4 1.6 1600 1650 1700 1750 1800 1850 1900 1950 2000 Year Resi dua l Ri ng-w id th South Flat
10 per. Mov. Avg. (South Flat)
(c)
correlated (r = 0.568), the White and Surprise sites are only slightly less correlated (r = 0.539; Table 2.2). Given the strong similarities between the chronologies, they were combined to form a regional master chronology with 87 cross dated series spanning 412 years (1596 to 2007; Figure 2.5). Five individual series did not significantly cross date and were removed. The EPS cutoff is at 1680, leaving the regional chronology with a mean inter‐series correlation of 0.544 and an average mean sensitivity of 0.244. Table 2.2: Correlation matrix showing interseries correlations between site ch nologies and master chronor o logy. S
urprise Wind Pay ss South Flat
Surprise ‐ 0.560 0.539 Windy Pass 0.560 ‐ 0.568 South Flat 0.539 0.568 ‐ 2.5.2 Climate station analysis The Stewart Airport and Dease Lake stations experience significantly different climates. At Stewart Airport the mean annual temperature is 5.6°C (± 3.5°C); whereas Dease Lake it averages ‐0.9°C (± 4.1°C; Figure 2.6). Average summer growing season air temperatures in June‐July‐August (JJA) are considerably different between the two stations, with conditions warmer at Stewart (mean JJA 14.3°C) than at Dease Lake (mean JJA 11.7°C).
Figure 2.5: Living mountain hemlock chronology displayed with a five‐year running mean. The number of series included in the chronology (sample depth) is shown at the bottom of the graph. The master chronology contains 27 series; the series intercorrelation value is 0.544 and the average mean sensitivity is 0.244.
Figure 2.6: A comparison of annual mean air temperature between Stewart Airport and Dease Lake. -4 -2 0 2 4 6 8 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year A nnua l Me a n T e mp (° C ) Stewart Airport Dease Lake Overall trends in the JJA air temperature show a 0.2°C warming in both instrumental data sets over the period of record (Figure 2.7). When analysing the overall trends, however, periods of warming and cooling are seen. At Stewart Airport, air temperatures warmed by 1°C from 1952 to 1978, followed by cooling of around 0.9°C from 1979 to 2001. The trend at Dease Lake is the opposite, with a ‐0.9°C cooling from 1958 to 1978, and 1.8°C of warming from 1979 to 2001. Figure 2.7 compares the annual mean precipitation at Stewart Airport and Dease Lake. At Stewart Airport total annual precipitation increased by 210 mm between 1910 and 2005. At Dease Lake, total annual precipitation increased by 25 mm between 1946 and 2003.
Figure 2.7: A comparison of annual mean air temperature between Stewart Airport and Dease Lake. 0 20 40 60 80 100 120 140 160 180 1940 1950 1960 1970 1980 1990 2000 2010 Year Pr eci pi tati on (mm) Stewart Airport Dease Lake
2.6 Analysis
2.6.1 ClimateGrowth Responses The master tree‐ring chronology was compared to mean monthly precipitation and monthly records of mean, minimum, and maximum temperature from both the Stewart and Dease Lake stations using DENDROCLIM2002. Correlations between ring width and climate variables from Stewart Airport and monthly precipitation totals at Dease Lake were all insignificant (Table 2.3). The chronology did, however, exhibit a strong positive correlation to the average mean JJA air temperature at Stewart (Figure 2.8) (r = 0.48 for the ring width chronology) and Dease Lake (r = 0.58 for the ring‐width chronology).Table 2.3: Correlations between instrumental climate variables and the residual
regional chronology. Significance level = 0.01.
Variable Stewart Airport1 Dease Lake
Minimum July Temperature 0.24 0.37 Maximum July Temperature 0.26 0.36 Mean July Temperature 0.14 0.49 Mean June‐July‐August Temperature 0.48 0.58 Total Monthly Precipitation 0.11 0.24 Figure 2.8: Coefficients calculated between the mountain hemlock chronology and the mean temperature time series from Dease Lake using DENDROCLIM2002. The lack bars represent correlation coefficients; the gray bars represent response unction coefficients. Months in block letters are from the previous year. b f
The climate‐radial growth scatterplot suggests a linear relationship exists between tree‐ring growth and mean JJA air temperature at Dease Lake (r2 = 0.34; Figure 2.9). An evolutionary moving average analysis (48 year blocks) performed with DENDROCLIM2002 indicates this relationship is consistent through time (Figure 2.10). Figure 2.9: Scatterplot comparing the radial growth of mountain hemlock and average June‐July‐August temperature at Dease Lake over the instrumental record (1943 – 2003). R2 = 0.3375 10 10.5 11 11.5 12 12.5 13 13.5 14 0.6 0.7 0.8 0.9 1 1.1 1.2
Residual Ring Width Indicies
Me an J une -Ju ly -Au gus t Te mpe ra ture ( °C) 1.3 This climate‐radial growth relationship is consistent with previous studies (Smith and Laroque 1998; Gedalof and Smith 2001a). It suggests that warm air temperatures in the JJA growing season lead to increased radial growth in mountain hemlock trees in this region (Figure 2.9).
Figure 2.10: This figure illustrates bootstrapped moving interval response function analyses calculated using the program DENDROCLIM2002. Forty‐eight year intervals were employed to test the strength of relationships between tree ring indices and the Dease Lake monthly June to August air temperature values over time. The panels depict the relationships between the mountain hemlock chronology and mean temperature values. Statistically significant positive elationships are highlighted in shades of red and negative relationships in shades f blue. Months of the previous year are identified with capital letters. r o 2.6.2 Dendroclimate Reconstruction Linear regression was used to model the relationship between mean JJA air temperature and radial growth for the period from 1944 to 2003, with split‐ verification partitioning the data into a 28 year calibration period from 1974 to 2003 and a 30 year verification period from 1944 to 1974 (Fritts 1976). The
modeled temperatures significantly correlate with the instrumental data (r = 0.63; Table 2.4), with the model successfully tracking low‐frequency variations in both the calibration and the verification periods (Figure 2.11). Given the strength of the modelled relationship, a proxy record of mean JJA air temperature was constructed extending to 1680 (Figure 2.12). Over the period of record the coolest summers occurred in 1706, 1810, 1880, 1970 and 1976. The warmest summers occurred in 792, 1804, 1816, 1885 and 1915. 1 Figure 2.11: June‐July‐August temperature model based on a linear relationship with the radial growth of mountain hemlock. 10 10.5 11 11.5 12 12.5 13 13.5 14 1940 1950 1960 1970 1980 1990 2000 2010 Year JJ A Te mp Instrumental Verification Calibration
for the Cambria Icefield. Figure 2.12: Reconstructed precipitation anomaly record The thick black line represents a 10 year running mean. 2.6.3 PDO and ENSO Circulation Patterns The master tree‐ring chronology correlates with SST in the Northeast Pacific and South Pacific oceans from 1870 to 2007 (Figure 2.13). Positive correlations (r = 0.3) occur across an east‐west equatorial belt centred at 170° east and terminating off the northwest coast of South America. These spatial relationships are interpreted as reflecting a radial growth response to warmer‐ than‐normal temperatures during positive ENSO phases. A positive relationship (r = 0.4) exist between radial growth and SST between 42° to 63° N offshore of PNA. The persistence and strength of these correlations suggest that a causal relationship exists between
Figure 2.13: Tree ring index – June‐July‐August sea surface temperature correlation. The tree rings have a positive relationship (r = 0.2 to 0.3) to sea surface temperatures across the equator, along the South American coast to 30˚ south, and off the coast of northern BC and Alaska. There is a negative correlation (r = ‐0.2 to ‐ 0.4) with sea surface temperature in the central north Pacific. radial growth and positive PDO phases that bring above normal summer growing season temperatures to this region (Mantua et al. 1997). The wavelet analysis aims to decompose a time series into time and space simultaneously and highlights both low and high frequency variability within the reconstructed proxy temperature record (Torrence and Compo 1998). Figure 2.14 indicates frequent 8‐year period events (marked in red) from 1600‐1700‐1820, 1870‐1900 and 1960‐1980, typical of ENSO, and weaker (shown in yellow) 32‐year period events from 1680‐1720, consistent with the significant variance of PDO
behaviour (Mantua and Hare 2002). A weakening of the PDO influence from 1800 and 1900 matches PDO reconstructions over this time period (Gedalof and Smith 001b). 2 Figure 2.14: (a) Summer Temperature Anomalies. (b) The wavelet power spectrum. The contour levels are chosen so that 75%, 50%, 25%, and 5% of the wavelet power is above each level, respectively. Black contour is the 5% significance level, using a white‐noise background spectrum. (c) The global wavelet power spectrum (black line). The dashed line is the significance for the global wavelet spectrum, assuming the same significance level and background spectrum as in (b).
2.7 Discussion
The radial growth of mountain hemlock trees along the eastern slopes of the northern Coast Mountains appears to be influenced by mean JJA air temperature. This climate‐radial growth relationship is assumed to reflect either a physiological response to increased rates of photosynthesis during warm summers (i.e., Kramer and Kozlowski 1960) or, alternatively, an indirect response resulting from accelerated melting of the seasonal snowpack during warm summers. The outcome is an extension of the growing season and potentially an extension of the period over which cambium is produced. These findings are largely consistent with those of earlier dendroclimatological investigations in PNA (Smith and Laroque 1998; Gedalof and Smith 2001a). However, most previous research has highlighted a negative relationship between the radial growth of mountain hemlock trees and spring snowpack depth (Graumlich and Brubacker 1986; Peterson and Peterson 2001; Smith and Laroque 1998); as well as a positive relationship with summer precipitation (Brubaker 1990). The reduced importance of seasonal precipitation in this setting is noteworthy and is attributed to rainshadow effects that lead to relatively dry leeward slopes (Environment Canada 2009). The three forest stands sampled in this study are found close to the easternmost extent of mountain hemlocks in the region where they are almost certainly periodically impacted by cold continental temperatures (Means 1990). It is also possible that in this setting the physiological role played by the seasonalsnowpack is not as significant as in more maritime settings (Smith and Laroque 1998; Larocque and Smith 2005). If this is the case, it is hypothesized that cool summer temperatures significantly limit the radial growth of mountain hemlocks.
The mean JJA proxy air temperature record constructed in this study illustrates warmer and cooler periods that are synchronous with those developed using dendroclimatological methodologies at other locations in the central and northern Coast Mountains (i.e., Gedalof 2002; Larocque and Smith 2003). Across this region cooler than normal JJA air temperatures characterize the intervals between 1698 to 1706, 1876 to 1886, and 1970 to 1976 (Figure 2.12). Warmer than normal JJA air temperatures occurred from 1715 to 1726, 1800 to 1820, 1900 to 1920 and 1945 to 1955. These alternating warm‐cold temperature regimes are in step with phases changes in the PDO and highlight the effect of both ‘warm’ and ‘cold’ PDO phases changes on JJA air temperature in this region. Like many previous dendroclimatological reconstructions this proxy record fails to accurately model extreme values (Figure 2.12). As only a moderate percentage of growth in any given year can be attributed to a single climate variable (Fritts 1976), the model response may be attributed to the role other limiting factors to growth play in anonymously warm or cool summers (i.e., Ettl and Peterson 1995). Given that variations in summer precipitation and spring snowpack depth have been shown to impact the radial growth of mountain hemlocks in many regions of PNA, it may be that during certain years the direct role of temperature diminishes.
Glacial histories from the northern Coast Mountains suggest their behaviour reflects a shared mass balance response to the temperature trends recorded in the proxy reconstruction. Jackson et al. (2008) show that glaciers were advancing in the Todd Icefield area from 1746 to 1764 and from 1843 to 1899. These intervals coincide with periods of lower than average summer temperature (Figure 2.12), suggesting that positive mass balances may be attributed to colder summer climates at this time. Recent research has led to recognition that trees located in climatically‐ sensitive at high altitude and latitude sites sometimes display an inconsistent radial growth response to climate through time (Visser et al. 2010). This divergent behavior appears confined to recent decades and hence strongly suggests an anthropogenic cause (Cook et al. 2004). In this instance, however, a bootstrapped moving interval response function analysis shows that over the period of instrumental record, mountain hemlock radial growth has consistently positively responded to JJA air temperatures (Figure 2.10). This finding suggests that as the Stewart region receives large amounts of winter precipitation, two responses occur: the ground is insulated during the winter months and hence soil water content is not a limiting factor and warm spring air temperatures in the spring melt the snowpack. Hence shallower snowpacks melt earlier in the season, and therefore tree‐ring growth can occur over a longer period of time.
2.8 Summary
A 400‐year long mountain hemlock tree‐ring chronology was constructed from increment samples collected at three sites in the northern Coast Mountains. Response function analysis showed that over the period from 1946 to 2007, the annual increment of radial growth was correlated to mean June‐July‐August temperature. A linear model was constructed to represent this relationship over the duration of the well‐represented portion of the chronology. The proxy temperature anomaly record constructed as part of this study extends from 1680 to 2007. Represented within the reconstructed record of summer temperature were extended intervals of warmer and cooler periods that closely match those previously reconstructed. Analysis of the likely casual factors responsible for these long‐term variations in temperature showed that climate forcing events described by the PDO and ENSO indices have played a significant role in long‐term climate changes in this region. In summary, this study demonstrates how temperature‐sensitive trees can be used to reconstruct robust proxy climate records for remote settings in northwestern British Columbia. Notably, the reconstruction expands upon the large‐scale temperature reconstructions presented by Briffa et al. (1994) and Briffa et al. (2001), by providing a detailed regional reconstruction of treeline temperature anomalies over the last 300 years. Understanding of the causes and effects of climate changes in this setting is crucial for evaluating the linked long‐ term hydroclimatic and glaciological impacts.Chapter 3
Dendroglaciological reconstruction of late Holocene glacier
activity at White and South Flat Glaciers, Boundary Range,
northern British Columbia Coast Mountains
3.1 Introduction
The rapid retreat and downwasting of glaciers in Pacific North America (PNA) over the last few decades is exposing land surfaces and glacially‐killed trees that provide singular insights into periods of glacier activity during the Holocene epoch. Recent and ongoing investigations in this region show that glaciers repeatedly advanced and retreated over the last eight millennia (Calkin et al. 2001; Menounos et al. 2009; Wiles et al. 2008). These discoveries are challenging earlier findings that Holocene glacier expansion was largely restricted to the Neoglacial period (Denton and Stuiver 1966; Ryder and Thompson 1986). Recognition of the dynamic character of glacier activity in the Holocene suggests a greater number of climate shifts than is recorded in many paleorecords from the region (Walker and Pellatt 2003). To understand the frequency and significance of these shifts, high resolution, long‐term temporally significant records are required. Annually‐resolved tree‐ring records from living trees and subfossil samples provide an opportunity to place current glaciological changes into a longer term climatological context (Swetnam and Betancourt 1997). Our understanding of Holocene glacial history in the mountains of PNA is incomplete. There are regions where little or no research focused on this topic hasbeen completed (Menounos et al. 2009). In British Columbia (B.C.), one such location is the northern Coast Mountains where limited research indicates that dendroglaciological records spanning the Holocene are present (i.e., Ryder 1987; Clague and Matthewes 1996; Jackson et al. 2008). The goal of the research presented in this chapter was to describe the late Holocene history of two glaciers flowing from the Cambria Icefield (Figure 3.1). Reconnaissance investigations in July 2008 revealed fresh exposures of subfossil wood buried in and below glacial sedimentary units. Additional sites and detrital subfossil wood deposits were discovered during a return expedition to the site in July 2009. Dendroglaciological research methodologies were applied to describe the ice front behavior of these two glaciers and to consider the climatological context of their Holocene behaviour.
3.2 Research Background
3.2.1 Treerings and dendroglaciology Dendrochronological techniques can be applied to study and date the movement of glaciers (Luckman 1998; Smith and Lewis 2007a). Dendrochronology, or the study of tree‐rings, deals with the dating and study of annual growth rings (Fritts 1976). As the radial growth of trees in the midlatitudes is largely limited by annual variations in climate, tree‐ring records can be cross dated to provide chronological and temporally extensive paleoenvironmental records (Fritts 1976). Dendroglaciological methodologies use tree‐ring evidence to provide insights into prehistorical glacier activity in two ways. Living‐trees found growingFigure 3.1: Location of the Cambria Icefield and key glaciers within the vicinity
and referred to in the text.
on recently deposited glacier deposits and landforms, or that were scarred by glacier activity, can be absolutely dated to provide minimum ages of surface stability and ice movement (Luckman 1998). In the case of moraines, the age of the oldest tree found growing on the deposit highlights the point in time when a glacier began to downwaste and/or retreat in response to a negative mass balance shift (Watson and Luckman 2004). Detrital wood and in situ tree stumps buried within or below glacial sedimentary units provide insights into intervals of glacier expansion and retreat (Smith and Lewis 2007b). The kill dates of subfossil wood samples are established by cross dating to dated tree‐ring chronologies or are referenced to radiocarbon‐dated floating tree‐ring chronologies to provide relative insights into when they were overwhelmed by an advancing glacier. 3.2.2 Holocene Glacial Activity in the British Columbia Coast Mountains
